execute-type param addition in GkeCodeExecutor #4111

SHRUTI6991 · 2026-01-10T00:49:29Z

Description:

Link to an existing issue (if applicable):

This PR updates the GkeCodeExecutor to add a new param called executor_type which takes either job / sandbox. Currently GkeCodeExecutor can only execute python code, so the changes made also focusses on executing ONLY python code via sandbox client.

Testing Plan

Unit Tests:

Addition of new Unit tests to verify forking of code between sandbox vs job.

[yes] I have added or updated unit tests for my change.
[yes] All unit tests pass locally.

Please include a summary of passed pytest results.

-- Docs: https://docs.pytest.org/en/stable/how-to/capture-warnings.html
======================== 9 passed, 2 warnings in 2.72s =========================
Finished running tests!

Manual End-to-End (E2E) Tests:

Integration test by running a simple AI Agent: https://google.github.io/adk-docs/tools/google-cloud/gke-code-executor/ to write and execute Python code. The sandbox was connected via local tunnel for testing.

2026-01-10 00:46:28,105 - INFO - sandbox_client.py:159 - Creating SandboxClaim 'sandbox-claim-4d0f82e0' in namespace 'agent-sandbox-system' using template 'python-sandbox-template'...
2026-01-10 00:46:28,340 - INFO - sandbox_client.py:180 - Watching for Sandbox to become ready...
2026-01-10 00:46:39,284 - INFO - sandbox_client.py:208 - Sandbox sandbox-claim-4d0f82e0 is ready.
2026-01-10 00:46:39,285 - INFO - sandbox_client.py:243 - Starting Dev Mode tunnel: localhost:49605 -> svc/sandbox-router-svc:8080...
2026-01-10 00:46:39,288 - INFO - sandbox_client.py:259 - Waiting for port-forwarding to be ready...
2026-01-10 00:46:40,289 - INFO - sandbox_client.py:271 - Dev Mode ready. Tunneled to Router at http://127.0.0.1:49605
2026-01-10 00:46:40,790 - INFO - gke_code_executor.py:156 - Executing code in sandbox:

def fibonacci(n):
  """
  This function calculates the nth Fibonacci number.

  Args:
    n: The index of the Fibonacci number to calculate (starting from 0).

  Returns:
    The nth Fibonacci number.
  """
  if n <= 0:
    return 0
  elif n == 1:
    return 1
  else:
    a = 0
    b = 1
    for _ in range(2, n + 1):
      c = a + b
      a = b
      b = c
    return b

# Get the first fibonacci number (index 0)
first_fibonacci = fibonacci(0) 

# Print the first fibonacci number
print(first_fibonacci)

2026-01-10 00:46:40,958 - INFO - sandbox_client.py:464 - File 'script.py' uploaded successfully.
2026-01-10 00:46:41,031 - INFO - sandbox_client.py:356 - Stopping port-forwarding...
2026-01-10 00:46:41,034 - INFO - sandbox_client.py:368 - Deleting SandboxClaim: sandbox-claim-4d0f82e0

Output:

Checklist

[ yes] I have read the CONTRIBUTING.md document.
[yes] I have performed a self-review of my own code.
[yes] I have commented my code, particularly in hard-to-understand areas.
[yes] I have added tests that prove my fix is effective or that my feature works.
[yes] New and existing unit tests pass locally with my changes.
[yes] I have manually tested my changes end-to-end.
[yes] Any dependent changes have been merged and published in downstream modules.

Additional context

Add any other context or screenshots about the feature request here.

…tor.

gemini-code-assist · 2026-01-10T00:49:47Z

Summary of Changes

Hello @SHRUTI6991, I'm Gemini Code Assist¹! I'm currently reviewing this pull request and will post my feedback shortly. In the meantime, here's a summary to help you and other reviewers quickly get up to speed!

This pull request enhances the GkeCodeExecutor by introducing a new executor_type configuration, enabling a choice between traditional Kubernetes Job-based execution and a new, more isolated sandbox environment. This change provides greater flexibility and security for running code, initially focusing on Python code within the sandbox, and lays the groundwork for future expansion of execution methods.

Highlights

Flexible Execution: Introduced an executor_type parameter in GkeCodeExecutor allowing users to choose between "job" (Kubernetes Job) and "sandbox" for code execution.
Sandbox Integration: Integrated the agentic_sandbox client to enable secure, sandboxed execution of Python code, addressing the current limitation of GkeCodeExecutor.
Code Refactoring: The core execute_code method was refactored into _execute_in_sandbox and _execute_as_job to cleanly separate execution logic based on the chosen executor_type.
Expanded Test Coverage: New unit tests were added to specifically verify the correct routing of code execution based on the executor_type parameter, ensuring both sandbox and job paths are correctly handled.

🧠 New Feature in Public Preview: You can now enable Memory to help Gemini Code Assist learn from your team's feedback. This makes future code reviews more consistent and personalized to your project's style. Click here to enable Memory in your admin console.

Using Gemini Code Assist

The full guide for Gemini Code Assist can be found on our documentation page, here are some quick tips.

Invoking Gemini

You can request assistance from Gemini at any point by creating a comment using either /gemini <command> or @gemini-code-assist <command>. Below is a summary of the supported commands on the current page.

Feature	Command	Description
Code Review	`/gemini review`	Performs a code review for the current pull request in its current state.
Pull Request Summary	`/gemini summary`	Provides a summary of the current pull request in its current state.
Comment	@gemini-code-assist	Responds in comments when explicitly tagged, both in pull request comments and review comments.
Help	`/gemini help`	Displays a list of available commands.

Customization

To customize Gemini Code Assist for GitHub experience, repository maintainers can create a configuration file and/or provide a custom code review style guide (such as PEP-8 for Python) by creating and adding files to a .gemini/ folder in the base of the repository. Detailed instructions can be found here.

Limitations & Feedback

Gemini Code Assist may make mistakes. Please leave feedback on any instances where its feedback is incorrect or counter productive. You can react with 👍 and 👎 on @gemini-code-assist comments. If you're interested in giving your feedback about your experience with Gemini Code Assist for Github and other Google products, sign up here.

You can also get AI-powered code generation, chat, as well as code reviews directly in the IDE at no cost with the Gemini Code Assist IDE Extension.

Review the Privacy Notices, Generative AI Prohibited Use Policy, Terms of Service, and learn how to configure Gemini Code Assist in GitHub here. Gemini can make mistakes, so double check it and use code with caution. ↩

gemini-code-assist

Code Review

This pull request introduces a new execution path in GkeCodeExecutor to use agentic-sandbox, controlled by a new executor_type parameter. The changes are logical and the addition of unit tests for the new branching logic is great. My feedback focuses on improving robustness, configurability, and code quality. Key suggestions include pinning the new git dependency for build stability, making the sandbox template name configurable, ensuring consistent logger usage, fixing a potential type error, and making the executor type selection logic more robust.

pyproject.toml

src/google/adk/code_executors/gke_code_executor.py

vicentefb · 2026-01-14T22:24:25Z

src/google/adk/code_executors/gke_code_executor.py

  kubeconfig_context: str | None = None
+
+  # Sandbox constants
+  python_sandbox_template: str = "python-sandbox-template"


Should we allow python_sandbox_template to be passed into init? If we ever want to support different template (e.g., a data-science specific image), we'll need to make this configurable.

Maybe i'm missing some context or forgetting something but yeah, why are we hardcoding the template name ?

I hardcoded it because, currently we only support python runtime. If we provide any runtime which doesn't support python. This code might fail.

Agreed on allowing the user to override the value. Whatever template they use must be available in their cluster, so even with this value hard-coded it can fail if not properly set up.

That being said I'd recommend being able to override the value as part of the execute_code rather than the init method. That way the executor can be used with multiple different templates (assuming that each template is valid and available in the cluster).

tests/unittests/code_executors/test_gke_code_executor.py

vicentefb

The separation between the Job and Sandbox logic is clean. I have just a few comments.

igooch · 2026-01-15T01:05:02Z

src/google/adk/code_executors/gke_code_executor.py

  kubeconfig_context: str | None = None
+
+  # Sandbox constants
+  python_sandbox_template: str = "python-sandbox-template"


Agreed on allowing the user to override the value. Whatever template they use must be available in their cluster, so even with this value hard-coded it can fail if not properly set up.

That being said I'd recommend being able to override the value as part of the execute_code rather than the init method. That way the executor can be used with multiple different templates (assuming that each template is valid and available in the cluster).

igooch · 2026-01-15T03:38:42Z

src/google/adk/code_executors/gke_code_executor.py

 class GkeCodeExecutor(BaseCodeExecutor):
  """Executes Python code in a secure gVisor-sandboxed Pod on GKE.

  This executor securely runs code by dynamically creating a Kubernetes Job for


Could you modify this comment block to indicate the difference between Job vs Sandbox? In particular that Sandbox will need other resources available in the cluster (agent-sandbox controller, sandbox-template, client router and gateway, etc.) in order for this code to properly work.

The comment here doesn't need to be extensive. A more detailed update should be added to the docs https://github.com/google/adk-docs/blob/main/docs/tools/google-cloud/gke-code-executor.md after this PR is merged.

SHRUTI6991 added 2 commits January 9, 2026 23:44

Add initial implementation to take in executor_type in gke_code_execu…

0b28736

…tor.

Fix extension name.

cf9a092

adk-bot added the tools [Component] This issue is related to tools label Jan 10, 2026

gemini-code-assist bot reviewed Jan 10, 2026

View reviewed changes

Address gemini comments to fix the dependency.

d5351f3

SHRUTI6991 changed the title ~~Execute type/param addition in GkeCodeExecutor~~ execute-type param addition in GkeCodeExecutor Jan 12, 2026

vicentefb reviewed Jan 14, 2026

View reviewed changes

src/google/adk/code_executors/gke_code_executor.py Show resolved Hide resolved

vicentefb reviewed Jan 14, 2026

View reviewed changes

src/google/adk/code_executors/gke_code_executor.py Show resolved Hide resolved

vicentefb reviewed Jan 14, 2026

View reviewed changes

tests/unittests/code_executors/test_gke_code_executor.py Show resolved Hide resolved

vicentefb suggested changes Jan 14, 2026

View reviewed changes

Address comments to add logging and validation.

ac876c5

SHRUTI6991 marked this pull request as ready for review January 15, 2026 18:16

igooch reviewed Jan 16, 2026

View reviewed changes

Provide feedback

Saved searches

Use saved searches to filter your results more quickly

Uh oh!

execute-type param addition in GkeCodeExecutor #4111

execute-type param addition in GkeCodeExecutor #4111

SHRUTI6991 commented Jan 10, 2026 •

edited

Loading

Uh oh!

gemini-code-assist bot commented Jan 10, 2026

Uh oh!

gemini-code-assist bot left a comment

Uh oh!

Uh oh!

Uh oh!

Uh oh!

Uh oh!

Uh oh!

Uh oh!

Uh oh!

vicentefb Jan 14, 2026

Uh oh!

SHRUTI6991 Jan 15, 2026

Uh oh!

igooch Jan 15, 2026

Uh oh!

Uh oh!

vicentefb left a comment

Uh oh!

igooch Jan 15, 2026

Uh oh!

igooch Jan 15, 2026

Uh oh!

Reviewers

Assignees

Labels

Projects

Milestone

Development

Uh oh!

4 participants

execute-type param addition in GkeCodeExecutor #4111

Are you sure you want to change the base?

execute-type param addition in GkeCodeExecutor #4111

Conversation

SHRUTI6991 commented Jan 10, 2026 • edited Loading Uh oh! There was an error while loading. Please reload this page.

Uh oh!

Testing Plan

Checklist

Additional context

Uh oh!

gemini-code-assist bot commented Jan 10, 2026

Summary of Changes

Highlights

Footnotes

Uh oh!

gemini-code-assist bot left a comment

Choose a reason for hiding this comment

Code Review

Uh oh!

Uh oh!

Uh oh!

Uh oh!

Uh oh!

Uh oh!

Uh oh!

Uh oh!

vicentefb Jan 14, 2026

Choose a reason for hiding this comment

Uh oh!

SHRUTI6991 Jan 15, 2026

Choose a reason for hiding this comment

Uh oh!

igooch Jan 15, 2026

Choose a reason for hiding this comment

Uh oh!

Uh oh!

vicentefb left a comment

Choose a reason for hiding this comment

Uh oh!

igooch Jan 15, 2026

Choose a reason for hiding this comment

Uh oh!

igooch Jan 15, 2026

Choose a reason for hiding this comment

Uh oh!

Reviewers

Assignees

Labels

Projects

Milestone

Development

Uh oh!

4 participants

SHRUTI6991 commented Jan 10, 2026 •

edited

Loading